[arin-ppml] Permanent ULA support in home networking (Re: New (-02) version of IPv6 CPE Router draft is available for review)

[bmanning at vacation.karoshi.com]

	you know, this outcome (ULA == GUA) was predicted from the 
	begining of the ULA emergence.  and the many assurances that
	*never* would ULA's *ever*  become GUAs are hollow.

--bill


On Tue, Jul 22, 2008 at 06:15:04PM -0400, Alain Durand wrote:
> [Changing the title of this long thread to focus the discussion and trimming
> the distribution list]
> 
> We have a trade-off here to make between network stability and application
> complexity: either  modify apps to deal with renumbering (ie try
> reconnecting, avoid well know literal addresses, use some kind of service
> location,...) or to teach application when it is ok to use ULA vs GUA,
> especially in referrals.
> 
> My take is that the former is simpler and lead to less service calls. The
> later introduce the notion that apps have to be aware of the topology, which
> I found disturbing.
> 
>   - Alain.
> 
> 
> On 7/22/08 6:05 PM, "Stark, Barbara" <bs7652 at att.com> wrote:
> 
> > Please don't change the ULA requirement. I believe ULA is needed. And I agree
> > that apps need to be intelligent about the scope of the addresses they are
> > using. I do not want to see the ULA go away. It needs to be persistent and
> > always there, from my perspective.
> > Barbara
> > 
> > -----Original Message-----
> > From: owner-v6ops at ops.ietf.org [mailto:owner-v6ops at ops.ietf.org] On Behalf Of
> > Hemant Singh (shemant)
> > Sent: Tuesday, July 22, 2008 5:02 PM
> > To: Alain Durand; Ralph Droms (rdroms)
> > Cc: Mark Townsley (townsley); Jimmy Chuang (cchuang); Rimi Denis-Courmont;
> > v6ops at ops.ietf.org; Wes Beebee (wbeebee)
> > Subject: RE: New (-02) version of IPv6 CPE Router draft is available for
> > review
> > 
> > Alain,
> > 
> > Very sorry, I missed your "*with* address referral" phrase.  Thanks for
> > providing the example.  Here is the analysis:
> > 
> > Yes, if C tries to communicate with B using B's ULA for destination, C will
> > also slap its ULA on the packet src address.  Thereafter before the CPE Router
> > WAN interface egresses the packet, the router has to comply to ULA forwarding
> > rules.  As per section 4.3 of RFC4193, the CPE Router will drop the packet
> > (unless the router is explicitly configured for a route to destination ULA)
> > and send an ICMPv6 Destination Unreachable to C. Here is the text from the
> > RFC.
> > 
> > [Site border routers and firewalls should be configured to not forward
> >  any packets with Local IPv6 source or destination addresses outside
> >  of the site, unless they have been explicitly configured with routing
> >  information about specific /48 or longer Local IPv6 prefixes.]
> > 
> > I don't expect the CPE Router to be supporting a site connected to another ULA
> > site so the question of any configuration on the CPE Router for a neighbor ULA
> > site is out of the question.  Since C gets some error indication, the app then
> > needs to figure out fixes in its implementation.
> > 
> > Sorry, I don't see this as rustication to change the CPE Router permanent ULA.
> > Some brain-dead apps need fixing.  I need more people to speak up and give
> > their opinion.  I am still open to change.
> > 
> > Thanks.
> > 
> > Hemant
> > 
> > -----Original Message-----
> > From: Alain Durand [mailto:alain_durand at cable.comcast.com]
> > Sent: Tuesday, July 22, 2008 1:25 PM
> > To: Hemant Singh (shemant); Ralph Droms (rdroms)
> > Cc: Mark Townsley (townsley); Jimmy Chuang (cchuang); Rimi Denis-Courmont;
> > v6ops at ops.ietf.org; Wes Beebee (wbeebee)
> > Subject: Re: New (-02) version of IPv6 CPE Router draft is available for
> > review
> > 
> > Hemant,
> > 
> > You missed the phrase "*with* address referral" in my response.
> > 
> > Say A & B are inside their home and use ULA & GUA. C and D are within another
> > home and are also using ULA & GUA.
> > 
> > Now, A B C & D enter a 4 way communication where they initially exchange the
> > addresses of their pier.
> > If A passes C the ULA and GUA of B, C might prefer to use B's ULA because of
> > address selection rules and C->B communication would fail or worse, go
> > somewhere else.
> > 
> > There are multiple variants of this. The point is that you cannot expect apps
> > that passes addresses to be smart enough to know about ULA & GUA.
> > 
> > BTw, using DNS does not help at all if you include both ULA & GUA AAAAs in
> > your zone...
> > 
> >   - Alain.
> > 
> > 
> > On 7/22/08 1:17 PM, "Hemant Singh (shemant)" <shemant at cisco.com> wrote:
> > 
> >> Alain,
> >> 
> >> Sorry I don't understand.  If any node in the home using an ULA sends
> >> a packet out the WAN interface of the CPE Router, the src-addr of the
> >> packet used is the GUA before the packet heads out of the node
> >> because, as we said in our draft, GUA has larger scope.  So any
> >> multi-party host on the Internet sees only the GUA.  I will need a
> >> specific example to show me how multi-party communications will break
> >> down with ULA and GUA configured on an interface of any node in the
> >> home behind the CPE Router or if ULA and GUA is configured on the LAN
> >> Interface of the CPE Router.
> >> 
> >> Thanks.
> >> 
> >> Hemant
> >> 
> >> -----Original Message-----
> >> From: Alain Durand [mailto:alain_durand at cable.comcast.com]
> >> Sent: Tuesday, July 22, 2008 11:48 AM
> >> To: Hemant Singh (shemant); Ralph Droms (rdroms)
> >> Cc: Mark Townsley (townsley); Jimmy Chuang (cchuang); Rimi
> >> Denis-Courmont; v6ops at ops.ietf.org; Wes Beebee (wbeebee)
> >> Subject: Re: New (-02) version of IPv6 CPE Router draft is available
> >> for review
> >> 
> >> On 7/21/08 12:43 PM, "Hemant Singh (shemant)" <shemant at cisco.com> wrote:
> >> 
> >>> I have repeatedly said, I am not convinced the ULA gets appreciable
> >>> complexity into the CPE Router. Our section 5.5.1 has clearly
> >>> outlined any complexity and shown it's minimal.  The ULA fixes a very
> >>> common problem for the CPE Router which is configuring the router
> >>> without any SP access - the problem is not a corner case.
> >> 
> >> Hemant,
> >> 
> >> 2 party communications in the presence of mixed ULA & GUA work ok,
> >> given proper default address selection rules.
> >> 
> >> Multi-party communications *with* address referral do not work in the
> >> general case in such a mixed environment, regardless of default address
> >> selection.
> >> 
> >>   - Alain.
> >> 
> >> 
> > 
> > 
> > 
> > 
> > *****
> > 
> > The information transmitted is intended only for the person or entity to which
> > it is addressed and may contain confidential, proprietary, and/or privileged
> > material. Any review, retransmission, dissemination or other use of, or taking
> > of any action in reliance upon this information by persons or entities other
> > than the intended recipient is prohibited. If you received this in error,
> > please contact the sender and delete the material from all computers. GA621
> > 
> > 
> 
> 
>